Cette thèse est dédiée à l'étude expérimentale et numérique des phénomènes de condensation dans un pressuriseur de chaufferie nucléaire embarquée conçu par l'entreprise DCNS. Au sein du système est observé un écoulement diphasique avec changement d'état, descendant dans un tube vertical partiellement immergé dans du liquide. Le liquide sous-refroidi s'écoule dans le tube sous l'effet de la gravité en entrainant de la vapeur saturée qui se condense à son contact. Dans un premier temps, l'étude des phénomènes de condensation est réalisée sur un banc expérimental dont la conception, le dimensionnement, la réalisation et l'exploitation ont été effectués au cours de cette thèse. Le dispositif est une version simplifiée à l'échelle 1 du système réel. Cette étude expérimentale a permis d'identifier les régimes d'écoulements diphasiques présents dans le système et d'analyser l'impact de trois paramètres : le débit de liquide injecté, le sous-refroidissement de ce liquide à l'entrée du tube et la pression en sortie de tube. Par ailleurs, l'étude propose un modèle déterminant les conditions pour laquelle le tube d'essais est complètement noyé, ce qui minimise les phénomènes de condensation. Dans un second temps, une simulation numérique de l'écoulement diphasique est réalisée par CFD sur ANSYS Fluent à l'aide du modèle " Volume Of Fluid ". La prise en compte de la condensation de la vapeur au contact du liquide sous-refroidi a nécessité le développement, grâce aux fonctions utilisateur du logiciel, d'un modèle spécifique s'appuyant sur un modèle de la littérature. Les résultats obtenus ont permis de reproduire qualitativement le déversement du liquide observé expérimentalement en entrée de tube ainsi que la condensation de la phase vapeur au contact de la phase liquide. / This thesis focuses on condensation phenomena occurring in a patented system designed by DCNS Company. The system ensures pressure regulation of nuclear boiler rooms embarked in naval vessels. Within the system occurs a downward liquid/vapor flow with phase change in a partially submerged vertical pipe. Subcooled liquid flows into the pipe and induce a suction of saturated vapor which condense in contact with liquid phase. This study aim to understand the physical phenomena occurring in the system and to simulate these phenomena. First, the condensation phenomena are studied with a test loop which conception, dimensioning, realization and exploitation were realized during this study. The experimental study identified the flow regimes occurring in the system and analyzed the impact of three parameters: the liquid flow rate, the liquid subcooling and the pressure at the system outlet. This study also proposed a simple model which determine the necessary conditions to completely drown the pipe, which minimize condensation phenomena. Thereafter, the numerical simulation of the two-phase flow was realized by CFD on ANSYS Fluent with the "Volume of Fluid" model. The condensation simulation was achieved by applying a condensation model used in literature. This model was added to the numerical code with User-Defined Functions (UDF). The simulation reproduced the weir regime flow at the pipe entrance and the vapor condensation inside the pipe.

Uma nova metodologia para a medida não intrusiva da fração volumétrica e da área interfacial é proposta neste trabalho, com base em redes neurais para processar respostas obtidas de sinais acústicos. A distribuição geométrica das fases dentro do escoamento é mapeada pela velocidade local de propagação acústica, considerada na equação diferencial que governa o problema. Esta equação é resolvida numericamente pelo método de diferenças finitas com as condições de contorno reproduzindo a estratégia de pulso/eco. Um número significativo de distribuições das velocidades de propagação foi considerado na solução da equação diferencial para construir uma base de dados, da qual os parâmetros da rede podem ser ajustados. Especificamente, o modelo neural é construído para mapear características extraídas dos sinais obtidos de quatro sensores acústicos, localizados no contorno externo do domínio de sensoriamento, estimando a fração volumétrica e a área interfacial correspondentes. Estas características correspondem às amplitudes e aos tempos de chegada dos três maiores picos da onda acústica. Os resultados numéricos mostram que o modelo neural pode ser treinado em um tempo computacional razoável e é capaz de estimar os valores da fração volumétrica e da área interfacial dos exemplos do conjunto de teste. / A new methodology for measuring the volumetric fraction and interfacial area in two-phase flows is proposed in this work, based on neural network for processing the responses obtained from an acoustic interrogation signal. The geometrical distribution of the phases within the flow is mapped by the local acoustic propagation velocity which is considered in the governing differential equation. This equation is solved numerically by the finite difference method with boundary conditions reproducing the pulse/echo strategy. A significant number of propagation velocities distributions were considered in the solution of the differential equation in order to construct a database from which the neural model parameters could be adjusted. Specifically, the neural model is constructed to map the features extracted from the signals delivered by four acoustic sensors, placed on the external boundary of the sensing domain, into the corresponding volumetric fraction and interfacial area. These features correspond to the amplitudes and the times of arrival on the three first peaks of the acoustic wave. Numerical results showed that the neural model can be trained in a reasonable computational time and it is capable of estimating the values of the volumetric fraction and the interfacial area of examples of the set of test.

Les boucles diphasiques à pompage capillaire sont des systèmes performants de transfert thermique qui peuvent transporter, de manière passive, de très importantes quantités de chaleur sur de grandes distances : utilisant les phénomènes qui apparaissent lors de la vaporisation d'un liquide dans un corps poreux (la mèche) contenu dans l'évaporateur, ces dispositifs offrent l'avantage de ne pas utiliser d'organe mécanique de puissance pour mettre lefluide frigorigène en mouvement. Vu les performances dont ces systèmes ont fait preuve dans le domaine spatial, leur utilisation dans le domaine gravitaire est aujourd'hui sérieusement étudiée, en particulier dans le domaine ferroviaire. Le travail de cette thèse porte sur la description du comportement thermodynamique instationnaire global d'une boucle diphasique à pompage capillaire, utilisée par Alstom Transport pour refroidir ses composants d'électronique de puissance. La première partie de ce travail consiste en la modélisation du transfert de masse et de chaleur dans la mèche contenu dans l'évaporateur, qui est le composant clé de la boucle. Un modèle mathématique instationnaire 2D a été développé pour décrire l'écoulement diphasique avec changement de phase dans le milieu poreux. Les résultats numériques de ce modèle montrent la formation d'une poche de vapeur au sein de la mèche poreuse. Une étude approfondie est élaborée pour décrire la dynamique de la croissance de cette poche en fonction de plusieurs paramètres (géométrie, sous-refroidissement, température de saturation, porosité). La deuxième partie de ce travail consiste à coupler le modèle de l'évaporateur au reste de la boucle afin d'obtenir un modèle thermodynamique global avec une interface liquide/vapeur mobile dans la mèche poreuse. Ce modèle est validé en régime transitoire par une confrontation de ses résultats aux mesures obtenues lors d'une précédente campagne expérimentale. Le comportement de la boucle au cours de la phase de démarrage est ensuite étudié. Finalement, la réponse transitoire de la boucle globale est analysée, en portant un intérêt particulier au comportement de l'évaporateur soumis à un créneau de puissance appliqué, à la variation de la température de consigne du réservoir de contrôle et à la variation de la température de la source froide en contact avec le condenseur. / Capillary pumped loop is a two phase device that uses the phase change phenomena occurred in the porous wick to transport large heat loads over long distances (several meters) without the need of any mechanical pump. CPLs have been developed and successfully employed for the thermal control of satellites. Thanks to their heat transport capacity, their use is now seriously considered in the gravity field especially for cooling electronic devices of a railroad traction chain. The work of this thesis focuses on the description of the unsteady thermodynamic behavior of a capillary pumped loop used by Alstom Transport to cool its power electronics components. The first part of the thesis consists to study the key component of CPL: the capillary evaporator. A 2D unsteady mathematical model has been developed to describe the heat and mass transfer inside the porous wick of the CPL evaporator. The numerical results of this model show the formation of a vapor pocket inside the porous wick. The influences of evaporator geometry, liquid subcooling, saturation temperature and wick porosity on the dynamic growth of the vapor pocket are discussed in detail. The second part of this work consists of coupling the CPL evaporator model with other CPL components in order to obtain a global capillary pumped poop model with a mobile liquid/vapor interface in the porous wick. The proposed model is validated by comparing numerical simulations with experimental results obtained in a previous works realized at Alstom Transport. The CPL startup is then studied. Finally, the influences of the applied power, reservoir temperature and heat sink temperature on the dynamic response of the overall loop are analyzed, with a particular interest to the evaporator behavior.

Les objectifs ambitieux, fixés aux acteurs du secteur automobile par les pouvoirs publics, en matière d'émission de polluants et de gaz à effet de serre rendent aujourd'hui indispensable une compréhension plus fine de la combustion dans les moteurs. La simulation 3D aux grandes échelles (LES) représente une voie prometteuse pour répondre à ces enjeux. Elle permet l'étude de phénomènes transitoires complexes inaccessibles avec des moyens expérimentaux ou des méthodes de calculs traditionnelles de type RANS. Ce travail de thèse est une première étape vers la simulation LES de l'injection de carburant liquide dans les moteurs à piston. Il a consisté à adapter le code de calcul aux particularités physiques de l'injection directe, technologie qui se généralise actuellement à tous les types de moteurs à piston. Dans un premier temps, et afin de s'affranchir du calcul 3D complexe en sortie d'injecteur, une méthodologie originale, consistant à initier le calcul en aval de l'injecteur, est proposée et validée sur différents cas. Pour la simulation 3D, l'approche Eulérienne mésoscopique, à laquelle est ajouté un modèle d'interaction particules-particules, est utilisée pour simuler le spray. Les simulations ont été premièrement validées par comparaison expérimentale dans des conditions proches de l'injection Diesel. De plus, une étude sur la dynamique du spray a permis de mieux comprendre son évolution et de dégager des points communs avec un jet de gaz turbulent. Des simulations complémentaires ont également montré la prédictivité de la LES sur des injections Diesel réalistes. Enfin, un premier calcul moteur à injection directe a été réalisé et a permis de valider les développements réalisés dans le cadre de cette thèse. / Car manufacturers are facing increasingly severe regulations on pollutant emissions and fuel consumption. To respect these regulations, a better understanding of combustion processes is needed. Large Eddy Simulation (LES) is becoming a promising tool for such issues as it allows the study of complex unsteady phenomena which can not be analysed with RANS simulations or experiments. The present work is a step towards the LES of liquid injection in piston engines. The numerical code has been adapted to the specifications of Direct Injection which is more and more used in industry. Firstly, in order to avoid the difficulties linked to the 3D simulation of cavitation, primary break-up and turbulence in the near-nozzle region, an original methodology, based on an injector model, has been proposed. The idea is to initiate the spray physics downstream to the injector exit. Then LES 3D simulations of spray have been conducted using the Eulerian Mesoscopic approach extended to dense dispersed sprays by the addition of a particle-particle interactions model. The simulation results have been validated by comparison with experimental data in Diesel conditions with a low injection pressure. Furthermore a study on the spray dynamics has permitted to better understand its development and to find similarities with a turbulent gaseous jet. Additional simulations on realistic Diesel injection conditions have shown the good predictivity of LES in such cases. Finally, a first simulation of a Direct Injection Engine has been been carried out to assess the developments achieved in this work.

Les turbines aéronautiques doivent satisfaire à des normes d’émissions polluantes toujours en baisse. La qualité du mélange du carburant et de l’air dans la chambre de combustion est responsable de la formation de polluants nocifs pour l’environnement. La simulation aux grandes échelles (LES) permet d’étudier les mécanismes de mélanges turbulents de l’air et du carburant. La prise en compte de l’aspect liquide du carburant injecté devient nécessaire pour prédire correctement l’apparition de vapeur de carburant au sein du foyer. Le but de cette thèse est évaluer la fiabilité des simulations LES Euler-Euler dans une configuration complexe. Les processus d’injection, et d’évaporation du carburant liquide sont analysés et modélisés dans les simulations LES car ils pilotent la formation de vapeur de carburant. Les méthodes numériques pour résoudre les équations continues de la phase dispersée doivent permettre des simulations précises et robustes dans une configuration représentative d’une chambre de combustion. Les simulations présentées dans ces travaux reproduisent l’écoulement diphasique évaporant non-réactif du banc d’essai Mercato. Ce banc est équipé d’un système d’injection d’air vrillé et d’un atomiseur pressurisé-swirlé de kérosène typiques des foyers aéronautiques réels. Dans ces travaux, le modèle pour l’injection de liquide FIM-UR a été développé pour définir les conditions limites conduisant à un spray issu d’un atomiseur préssurisé-swirlé. Le kérosène employé dans les campagnes expérimentales est modélisé dans les simulations par un composé permettant d’obtenir des temps d’évaporation réalistes. Trois stratégies numériques ont été mises en place sur la configuration Mercato. Les comparaisons des résultats numériques aux mesures expérimentales ont permis d’évaluer la stratégie numérique conduisant à la meilleure précision. L’utilisation du schéma centré TTGC associé à un opérateur de viscosité artificielle localisée par un senseur adapté est optimale lorsque l’équation sur l’énergie décorrélée des gouttes est résolue. Cette stratégie permet de contrôler la localisation et les niveaux de viscosité par rapport à un schéma décentré. Les termes sources liés au mouvement mésoscopique permettent de redistribuer l’énergie dans les zones de compression ou de détente de la phase dispersée, et d’obtenir les bonnes répartitions des fluctuations dans la chambre de combustion. La stratégie retenue est comparée aux statistiques de la dynamique du spray résolu par une approche Lagrangienne employant la même injection monodispersse. Le méthode Euler-Euler conduit à la même précision de la dynamique de la phase dispersée que la méthode Euler-Lagrange. L’accès à l’évolution instationnaire de l’écoulement permet d’identifier les mêmes mécanismes de dispersion et de mélange dans les deux simulations. Des différences sur la répartition de diamètre moyen et de carburant dans la chambre ont été mis en évidence et reliés à la polydispersion locale qui n’est pas résolue dans l’approche Euler-Euler monodisperse et qui apparaît naturellement dans l’approche Euler-Lagrange malgré l’injection monodisperse. / Aeronautical gas turbines are facing growing demands on emission reductions. Indeed, the quality of the air-fuel mixture directly triggers the formation of pollutants degrading the environment. Large Eddy Simulation is an accurate numerical method to predict turbulent mixing in combustors. Adding the liquid phase of the fuel in these simulations also becomes necessary to properly predict the injection process and the vaporization of the fuel in the combustion chamber. The purpose of this dissertation is to evaluate the accuracy and reliability of Euler-Euler LES in a complex combustor configuration. The injection and vaporization processes of the fuel liquid phase are both modeled in the present LES as they drive the formation of the fuel gas phase. Moreover, the numerical methods that solve the continuous equations of the disperse phase must be accurate and robust in realistic combustor configurations. The simulations shown in the present study reproduce the non-reactive two-phase flow of the ONERA Mercato test bench. The experimental set-up is equipped with an air-swirler injection system and a pressure-swirled atomizer typical of actual turboengine combustors. In the present work the FIM-UR liquid injection model has been developed. It creates boundary conditions profiles for a liquid spray produced by a pressure-swirled atomizer. Kerosene used in the experiments is modeled in the present numerical simulations by a single species leading to a good estimate of the vaporization rate. Three numerical strategies have been tested on the Mercato configuration. Comparisons between experimental and LES results help defining the most accurate numerical strategy. The use of the centered numerical scheme TTGC stabilized by a localized artificial viscosity operator is best when the random uncorrelated energy of droplets is also resolved. Unlike an upwind numerical scheme, the selected strategy allows the user to control where and how much artificial viscosity is added. The source terms coming from the mesoscopic movement redistribute the energy in the compression or expanding zones of the disperse phase, and provide the proper distribution of fluctuations in the combustion chamber. The obtained strategy is compared with the statistics provided by a Lagrangian description of the liquid spray in the same mono-disperse injection. The Euler-Euler approach leads to the same accuracy in the same spray dynamics of the disperse phase as in the Euler-Lagrange method. Both unsteady flow simulations also provide the same dispersion and mixing processes in the Mercato set-up. Differences on the mean diameter and the fuel distribution in the combustion chamber are seen and related to the local poly-dispersion that cannot be resolved in the mono-disperse Euler-Euler approach and that naturally appear in the Euler-Lagrange method despite the mono-disperse injection.

Um dos principais problemas relacionados ao transporte e manipulação de fluidos multifásicos refere-se a existência de regimes de escoamento e sua forte influência sobre importantes parâmetros de operação. Um bom exemplo disto ocorre em reatores químicos gás-líquido no qual máximos coeficientes de reação podem ser alcançados mantendo-se um escoamento a bolhas disperso, maximizando a área interfacial total. Assim, a habilidade de identificar automaticamente regimes de escoamento é de importância crucial, especialmente para operação adequada de sistemas multifásicos. Este trabalho se constituirá no desenvolvimento e implementação de uma rede neural auto-organizativa especializada ao problema de identificação de regimes de escoamento bifásico ar-água em tubo horizontal. Os regimes reconhecidos em tubo horizontal são seis: estratificado liso, estratificado ondulado, estratificado rugoso, intermitente, bolhas e anular. Para tanto, pretende-se tomar como ponto de partida as medidas capacitivas, variação de pressão da tubulação e medida de pressão \"flutuante\" como padrões associativos a serem armazenados na rede neural. Posteriormente, a fase de treinamento da rede neural consistirá em identificar os coeficientes sinápticos apropriados, a partir de um conjunto representativo de ensaios. Nesse trabalho optou-se por uma arquitetura com 10 neurônios na camada de entrada, portanto uma quantidade maior do que o número de regimes que se deseja identificar. O objetivo é ver se a rede neural consegue encontrar de forma autônoma os seis regimes já conhecidos, mesmo tendo 10 neurônios na grade. Isso demonstra a habilidade da rede neural auto-organizativa em identificar regimes de escoamento mesmo em situações onde não há conhecimento prévio dos mesmos. Serão feitas simulações para verificar o desempenho da rede neural a partir de dados experimentais coletados no oleoduto piloto do Núcleo de Engenharia Térmica e Fluidos da Escola de Engenharia de São Carlos. / One of the main problems related to the transport and manipulation of multiphase flow is the existence of flow regimes and its strong influence on important parameters of operation. As an example of it occurs in gas-liquid chemical reactor in which maximum coefficients of reaction can be reached by keeping a maximal flow bubbly dispersed into a total interfacial area. Thus, the ability to identify flow regimes automatically is very important, specially in order to obtain a properly operation of multiphase systems. This work consists in the development and implementation of a self-organizing neural networks specially for the problem of identification of regimes of two-phase flow air-water in horizontal pipe. The regimes recognized in horizontal pipe are six: estratified smooth, estratified wavy, estratified rugged, intermittent, bubbly and annular. The capacitive measures, variation of pressure of the piping and measure of flutuante pressure were considered as a starting point to be stored as associative standards in the neural networks. After, the phase of training of the neural networks consisted of an appropriate identification for the sinaptic coefficients from a representative set of tests. An architectural input-layer of 10 neurons was opted. Therefore a bigger amount of regimes than the ones usually required. The objective is to see if the neural network is able to find independently the six regimes already known, even having 10 neurons in the grid. This demonstrates the ability of the self-organizing maps neural network in identifying flow regimes in situations where one does not have previous knowledge of them. Simulations will be made to verify the performance of the neural network from experimental data collected in the pilot pipe-line of the Nucleus of Thermal Engineering and Fluids of the School of Engineering of São Carlos.

[pt] Esse trabalho apresenta um modelo numérico para a análise de processos acoplados de efeitos mecânicos e escoamento bifásico em meios porosos fraturados, com a utilização de diferentes métodos numéricos combinados (Elementos finitos contínuos e descontínuos), e utilizando uma mesma malha de elementos finitos para representar uma célula de modelo de simulação de reservatório. As descontinuidades são inseridas na malha como elementos de nós duplicados colapsados. Empregam-se procedimentos numéricos desenvolvidos em dois grupos distintos. Um primeiro grupo de simulações trata de um procedimento numérico de escoamento bifásico, com ênfase à obtenção de um balanço de volumes verdadeiramente conservativo. Nesta fase, avalia-se uma formulação numérica que emprega um processo em três etapas: o método dos elementos finitos (EF), para a aproximação da equação da pressão; intermediariamente, utiliza-se o método de Raviart-Thomas de mais baixa ordem para aproximação da velocidade; e a aproximação da equação da saturação pelo método dos elementos finitos descontínuos (MEFD), que utiliza um limitador de inclinação multidimensional de modo a evitar oscilações na reconstrução dos dados de saturação. Para fins de validação da formulação desenvolvida, comparam-se os resultados obtidos com simulações utilizando o Método do Volumes Finitos (VF). O segundo grupo de simulações trata de acoplar o módulo mecânico (em EF) ao simulador de escoamento bifásico, de modo que a variação do estado de tensões, decorrente da explotação, seja considerada nas simulações. Essa análise permite uma melhor representação do fenômeno estudado além de proporcionar melhores resultados quanto à chegada da água nos poços produtores, afetando a produtividade do modelo. / [en] This work presents a numerical model for the analysis of coupled processes of mechanical effects and two-phase flow in fractured porous media using different numerical methods (continuous and discontinuous finite elements), and using the same finite element mesh to represent a cell of reservoir simulation model. The discontinuities are inserted into the mesh as elements of collapsed duplicate nodes. Numerical procedures developed in two distinct groups are used. A first group of simulations deals with a numerical two-phase flow procedure, with special emphasis on obtaining a truly conservative volume balance. At this stage, a numerical formulation using a three-step process is evaluated: The Finite Element Method (EF), for the approximation of the pressure equation; the lower order of Raviart-Thomas method is used to velocity approximation; and the approximation of the saturation equation by the discontinuous finite element method (MEFD), using a multidimensional slope limiter in order to avoid oscillations in the reconstruction of the saturation data. For the validation of the developed formulation, the results obtained are compared with simulations using the Finite Volume Method (VF), with focus on the analysis of the conservation of volumes. The second group of simulations couple the mechanical module (in EF) to twophase flow computer program, so the variation of the stress state, due to exploitation, is considered in the simulations. This analysis allows a better representation of the phenomenon. In addition to providing better results regarding the arrival of water in the producing wells, affecting the productivity of the model.

Um dos principais problemas relacionados ao transporte e manipulação de fluidos multifásicos refere-se a existência de regimes de escoamento e sua forte influência sobre importantes parâmetros de operação. Um bom exemplo disto ocorre em reatores químicos gás-líquido no qual máximos coeficientes de reação podem ser alcançados mantendo-se um escoamento a bolhas disperso, maximizando a área interfacial total. Assim, a habilidade de identificar automaticamente regimes de escoamento é de importância crucial, especialmente para operação adequada de sistemas multifásicos. Este trabalho se constituirá no desenvolvimento e implementação de uma rede neural auto-organizativa especializada ao problema de identificação de regimes de escoamento bifásico ar-água em tubo horizontal. Os regimes reconhecidos em tubo horizontal são seis: estratificado liso, estratificado ondulado, estratificado rugoso, intermitente, bolhas e anular. Para tanto, pretende-se tomar como ponto de partida as medidas capacitivas, variação de pressão da tubulação e medida de pressão \"flutuante\" como padrões associativos a serem armazenados na rede neural. Posteriormente, a fase de treinamento da rede neural consistirá em identificar os coeficientes sinápticos apropriados, a partir de um conjunto representativo de ensaios. Nesse trabalho optou-se por uma arquitetura com 10 neurônios na camada de entrada, portanto uma quantidade maior do que o número de regimes que se deseja identificar. O objetivo é ver se a rede neural consegue encontrar de forma autônoma os seis regimes já conhecidos, mesmo tendo 10 neurônios na grade. Isso demonstra a habilidade da rede neural auto-organizativa em identificar regimes de escoamento mesmo em situações onde não há conhecimento prévio dos mesmos. Serão feitas simulações para verificar o desempenho da rede neural a partir de dados experimentais coletados no oleoduto piloto do Núcleo de Engenharia Térmica e Fluidos da Escola de Engenharia de São Carlos. / One of the main problems related to the transport and manipulation of multiphase flow is the existence of flow regimes and its strong influence on important parameters of operation. As an example of it occurs in gas-liquid chemical reactor in which maximum coefficients of reaction can be reached by keeping a maximal flow bubbly dispersed into a total interfacial area. Thus, the ability to identify flow regimes automatically is very important, specially in order to obtain a properly operation of multiphase systems. This work consists in the development and implementation of a self-organizing neural networks specially for the problem of identification of regimes of two-phase flow air-water in horizontal pipe. The regimes recognized in horizontal pipe are six: estratified smooth, estratified wavy, estratified rugged, intermittent, bubbly and annular. The capacitive measures, variation of pressure of the piping and measure of flutuante pressure were considered as a starting point to be stored as associative standards in the neural networks. After, the phase of training of the neural networks consisted of an appropriate identification for the sinaptic coefficients from a representative set of tests. An architectural input-layer of 10 neurons was opted. Therefore a bigger amount of regimes than the ones usually required. The objective is to see if the neural network is able to find independently the six regimes already known, even having 10 neurons in the grid. This demonstrates the ability of the self-organizing maps neural network in identifying flow regimes in situations where one does not have previous knowledge of them. Simulations will be made to verify the performance of the neural network from experimental data collected in the pilot pipe-line of the Nucleus of Thermal Engineering and Fluids of the School of Engineering of São Carlos.

Este trabalho tem como objetivo global contribuir para o desenvolvimento do estudo da propagação de ondas em dutos de transporte de fluidos, especialmente oleodutos e gasodutos, através da análise da atenuação acústica e da velocidade de propagação da onda. Para isso, foram feitos ensaios experimentais, tanto em escoamento mono quanto bifásico, variando-se a vazão do escoamento em uma linha de testes de 1500 metros e 50 mm de diâmetro, para aquisição de sinais gerados a partir do fechamento de válvulas de simulação de vazamentos em pontos ao longo da tubulação. Para análise da atenuação acústica foi proposta uma metodologia baseada na medição do coeficiente de amortecimento temporal das ondas de pressão como subsídio para o cálculo do coeficiente de atenuação, através da utilização de um único sensor de pressão. Esta metodologia foi validada pela medição direta do coeficiente de atenuação e os resultados mostram uma excelente concordância em escoamentos monofásicos. Para os testes em escoamento bifásico não foi possível aplicar esta metodologia devido ao forte amortecimento das ondas de pressão, sendo feitas apenas medições diretas da atenuação com resultados fenomenologicamente coerentes. A velocidade de propagação acústica para escoamento monofásico foi influenciada pela presença de gás dissolvido e bolhas de cavitação, apresentando valores próximos aos da literatura para baixas vazões de escoamento. Para altas vazões, as velocidades foram compatíveis com o escoamento bifásico a bolhas finamente dispersas. Para o escoamento bifásico o regime de escoamento variou de bolhas a pistonado. As velocidades de propagação obtidas são compatíveis com aquelas obtidas por modelos teóricos. / This work aims to contribute to the development of the study of wave propagation in fluid transport pipelines, through the analysis of acoustic attenuation and propagation speed. For this reason, experimental tests were made in single and two-phase flow, varying the flow rate in a test line of 1500 meters and 50 mm in diameter, for the acquisition of signals generated from the closing of leak simulation valves at known points along the pipeline. To analyze the acoustic attenuation it was proposed a methodology based on measurement of temporal damping coefficient of pressure waves as input for calculating the attenuation coefficient, by using a single pressure sensor. This methodology was validated by direct measurement of attenuation coefficient and the results show an excellent agreement in single-phase flow. For the tests in two-phase flow it was not possible to apply this methodology due to the strong damping of pressure waves, which only made direct measurements of attenuation results phenomenologically consistent. Acoustic propagation velocity for single phase flow was influenced by the presence of dissolved gas and cavitation bubbles, having values close to the literature for low flow rates. For high flows rates, the velocities were consistent with a two-phase flow at finely dispersed bubbles. For the two-phase flow the regime varied from slug to bubbles. The propagation velocities obtained are consistent with those obtained by theoretical models.

Cette étude concerne l’étude des écoulements diphasiques dans le cadre du refroidissement des composants électroniques en systèmes embarqués. L’étude bibliographique a permis de sélectionner le refroidissement par spray comme technique prometteuse pour dissiper des flux de chaleur au-delà de 100 W/cm2. Une étude hydraulique, utilisant de l’eau et du HFE7100 comme fluides de refroidissement, nous a permis de valider des modèles permettant de déterminer la taille et la vitesse de gouttes provenant d’une sélection de buses de spray. Pour la partie thermique, nous avons conçu une section d’essais (évaporateur) permettant de pulvériser en spray afin d’étudier le refroidissement avec une boucle fermée diphasique.Vu la complexité du système de spray influencé par plusieurs paramètres et phénomènes physiques, nous avons isolé le phénomène d’ébullition nucléée dans une configuration en ébullition nucléée avec un élément chauffant identique à celui employé avec le refroidissement par spray. Pour améliorer les échanges thermiques, 6 surfaces avec différentes structurations (macroscopiques, microscopiques et hybrides) ont été sélectionnées. Les résultats de tests avec ces surfaces ont été comparés avec une surface lisse tant pour le refroidissement par spray que pour le refroidissement en vase.D’une part, avec un refroidissement par spray, les surfaces macrostructurées nous ont permis de dissiper des puissances thermiques de l’ordre de 140 W/cm2 avec d’importants coefficients de transfert thermique. D’autre part, avec un système de refroidissement par immersion, une des surfaces hybrides a montré être la plus performante.Les résultats reportés dans cette thèse ont permis d’approfondir la compréhension des mécanismes de transfert de chaleur en refroidissement par spray. De même, ils ouvrent la voie à l’étude des améliorations et optimisations du système permettant de l’employer en systèmes embarqués. / This dissertation concerns the study of two-phase flow cooling of electronic components in embedded systems. From a literature review, Spray Cooling was selected as a promising technique for dissipating heat fluxes above 100 W/cm2. A hydraulic study, using water and HFE7100 as coolants, has validate models for determining the size and speed of drops from a selection of spray nozzles. Regarding the thermal study, we have designed a test section (evaporator) to study cooling in a two-phase closed loop system.Given the complexity of Spray Cooling systems, which are influenced by several parameters and involve several physical phenomena, the nucleate boiling phenomenon has been isolated in a Pool Boling system with an identical heating element as Spray Cooling experiment. To improve heat exchange, 6 surfaces with different structures (macroscopic, microscopic and hybrid) were selected. The boiling test results with these surfaces have been compared with a smooth surface for both Spray Cooling and Pool Boiling.On one hand, in Spray Cooling tests, the macrostructured surfaces dissipated heat flux up to 140 W/cm2 with significant heat transfer coefficients. On the other hand, in the Pool Boling system, one of the hybrid surfaces has shown to be the most efficient.The results reported in this dissertation contributes on the understanding of the boiling mechanisms of heat transfer in Spray Cooling. Likewise, they open the way to the study of improvements and optimizations of the system for its use in embedded systems.